Skip to content

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

28 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Agent Skills

Multi-agent evaluation systems, persistent error memory, C-to-Rust migration, composable pipelines, and browser automation for AI coding assistants.

28 open-source skills. Works with Claude Code, Codex CLI, Gemini CLI, Cursor, and Windsurf.

HACKATHON JURY SIMULATOR — ORE at Colosseum Renaissance
5 jurors based on real Colosseum judges | Grand Prize Winner (8,300 participants)

| Criterion (Weight)  | Matty T. | Nate L. | Mert M. | Clay R. | Chase B. |  Avg |
|---------------------|----------|---------|---------|---------|----------|------|
| Innovation (25%)    |    9     |    9    |    9    |    9    |    8     |  8.8 |
| Technical (25%)     |    7     |    9    |    9    |    7    |    8     |  8.0 |
| Impact (20%)        |   10     |    8    |    8    |    9    |   10     |  9.0 |
| UX/Design (15%)     |    7     |    7    |    6    |    8    |    7     |  7.0 |
| Presentation (15%)  |    9     |    7    |    8    |    9    |    8     |  8.2 |
| Weighted Total      |  8.40    |  8.20   |  8.20   |  8.35   |  8.25    | 8.28 |

"ORE spawned an entire sub-ecosystem: pools, CLIs, boost protocols,
 the steel framework. That is the definition of ecosystem contribution."
 — Chase Barker, Solana Foundation

Skills

Skill Description Category
architect Multidisciplinary idea and project evaluator. Dispatches 3-6 specialist evaluators in parallel, synthesizes into unified assessment with scores, Value Matrix, and next steps. Multi-Agent Panels
council Life advisory council with 8 parallel specialist advisors and synthesis. Covers finance, health, career, learning, systems, relationships, and creative. Multi-Agent Panels
full-eval Complete strategic and personal evaluation. Runs architect and council in parallel, cross-references outputs, generates PDF report. Multi-Agent Panels
hackathon-jury Simulate a real hackathon jury. Researches actual jury members, builds personalized juror personas, dispatches parallel evaluations. Multi-Agent Panels
vc-evaluator Simulate a VC investment panel. Researches real investors, builds personas from their thesis and portfolio, synthesizes partner meeting deliberation. Multi-Agent Panels
oss-readiness Evaluate open source readiness. 5 specialist agents audit licensing, docs, security, community, and API surface in parallel. Multi-Agent Panels
founder-mode End-to-end product builder with founder mindset. Orchestrates 9 phases from idea evaluation to launch with README, pitch, and Product Hunt draft. Multi-Agent Panels
idea-lab Automated idea laboratory. Generates diverse ideas, pre-screens them, evaluates best with multi-agent scoring, ranks winners. Multi-Agent Panels
c-to-rust-migration Complete C/C++ to Rust migration pipeline. Analyzes difficulty, converts to idiomatic safe Rust, validates results. Rust Migration
c-cpp-analyzer Analyze C/C++ code for Rust migration planning. Classifies functions by difficulty, identifies unsafe patterns and pointer complexity. Rust Migration
rust-migration C/C++ to idiomatic Rust conversion patterns. Covers pointer translation, memory management, error handling, and quality gates. Rust Migration
migration-validator Safety verification for C-to-Rust migrations. Validates compilation, unsafe minimization, memory safety, and API compatibility. Rust Migration
noricum-dev Development conventions for the Noricum C-to-Rust migration agent. Architecture decisions, crate structure, and coding standards. Rust Migration
demo-pipeline Record automated demos for mobile and web apps. Converts natural language scripts into Maestro flows or webreel configs. Testing & Recording
e2e-pipeline Generate and run E2E tests for mobile and web. Discovers screens and flows, generates Maestro test flows, reports results. Testing & Recording
maestro-mobile-testing Maestro mobile E2E testing patterns for React Native/Expo. YAML test flows, adaptive auth, GraalJS scripting, CI/CD integration. Testing & Recording
webreel Create scripted browser demo videos. Generates MP4, GIF, or WebM with cursor animation, keystroke overlays, and sound effects. Testing & Recording
agent-browser Browser automation for AI agents via inference.sh. Navigate, interact, screenshot, and record video from web pages. Testing & Recording
playwright-recording Record browser interactions as video using Playwright. Cursor highlighting, click ripple effects, and window scaling. Testing & Recording
pdf-presentation Generate professional PDF reports and presentations. Tables, score visualizations, and styled sections from structured content. Design
mobile-design-system Comprehensive mobile UI/UX design system. React Native components, touch interactions, gesture patterns, and accessibility. Design
ios-glass-ui Redesign mobile app UI with modern Apple-like glass materials. Translucency, blur, depth, and SF Pro typography. Design
app-store-screenshots Create App Store and Google Play screenshots with exact platform specs. Device mockups, captions, and A/B testing. Design
ui-ux-polish Iterative UI/UX polishing workflow. Achieve Stripe-level visual polish through multiple incremental passes. Design
learn-by-mistake Persistent error memory. Analyzes failures, extracts root-cause lessons, prevents the same mistake twice. Analysis
post-run-analysis Automatic post-mortem analysis. Parses logs, compares with history, identifies recurring patterns, proposes fixes. Analysis
architecture-reviewer Audit code against architecture documentation. Detects drift between design docs and actual code structure. Analysis
youtube-devrel Create developer YouTube content and technical screencasts. Tutorial structure, live coding, thumbnails, and SEO. Development

Multi-Agent Panels

One command: 6 business evaluators + 8 life advisors + PDF report. Orchestrates architect, council, and pdf-presentation into a single end-to-end evaluation. You describe an idea, it dispatches 14 specialist agents in parallel, synthesizes two independent verdicts, resolves cross-system conflicts, and delivers a polished PDF. The complete strategic + personal evaluation in one shot. Requires: Agent tool, WebSearch

Web-searches the actual judges of your hackathon, builds 3-7 juror personas from their real backgrounds and biases, dispatches each to evaluate your project independently, then simulates the deliberation — scores, individual feedback, placement estimate, and pitch coaching. Nothing comparable exists. Requires: WebSearch, Agent tool

6 parallel specialist evaluators — market, technical, risk, impact, resources, innovation — each assess your idea independently. A synthesizer resolves conflicts and produces a unified Opportunity Score, Value Matrix, and Go/No-Go recommendation. Dual-mode: raw ideas from scratch or existing codebase health scans. Requires: Agent tool, WebSearch

Simulate a real VC investment panel. Researches actual investors in your sector, builds personas from their investment thesis and portfolio, runs parallel evaluations, and simulates the Monday partner meeting with funding verdict, feedback, and pitch coaching. Requires: WebSearch, Agent tool

Open source readiness evaluator — 5 parallel agents audit licensing, documentation, security, community readiness, and API surface. Produces a readiness score and prioritized checklist for going open source. Requires: Agent tool

8 parallel life advisors — finance, health, career, relationships, creativity, productivity, social dynamics, personal growth. Each evaluates your situation independently, then a synthesizer unifies their recommendations. The personal counterpart to architect's business analysis. Requires: Agent tool


Error Memory

Your AI never makes the same mistake twice. Analyzes every failure, extracts the root cause as a structured lesson, and persists it. Next time the same pattern appears, the fix is applied before the error can happen. Confidence gate (2 occurrences before promoting), secret sanitization, and automatic pruning built-in. Commands: /learn, /lessons, /forget. Requires: None (hooks optional)

Automatic post-mortem after iterative runs (migration, build, deploy, test suites). Parses logs, compares with previous runs, identifies recurring patterns, classifies findings as FIXABLE/NEEDS_INVESTIGATION, and updates project memory. The feedback loop closer. Requires: None


Architecture

Audit code against architecture documentation. Detects drift between what was designed and what was built — boundary violations, dependency drift, responsibility creep, technology changes. Generates drift reports with severity classification. Includes ARCHITECTURE.md starter template. Requires: None


Rust Migration

Complete C/C++ to Rust migration pipeline. Analyzes code difficulty (easy/medium/hard), converts to idiomatic safe Rust with proven patterns (malloc->Vec, error codes->Result, NULL->Option), validates with differential testing. Requires: Rust toolchain

Analyze C/C++ code for Rust migration planning. Classifies functions by migration difficulty, identifies unsafe patterns, pointer complexity, and macro usage. Requires: None

C/C++ to idiomatic Rust conversion patterns and pipeline. Covers pointer translation, memory management, error handling, string conversion, struct patterns, and quality gates. Requires: Rust toolchain

Safety verification checklist for C-to-Rust migrations. Validates compilation, unsafe block minimization, memory safety, API compatibility, and idiomatic Rust patterns. Requires: Rust toolchain

Development conventions for the Noricum C-to-Rust migration agent. Architecture decisions, crate structure, coding standards, and contribution guidelines. Requires: None


Ideation

Automated idea laboratory. Describe your context (hackathon, need, problem type), and it generates N diverse ideas using forced divergence axes (SCAMPER, cross-domain analogies), pre-screens them with a quick evaluation, deep-evaluates the top ones via /architect (6 parallel evaluators each), and presents a ranked dashboard with scenario-based recommendations. Composes with /founder-mode to build the winning idea end-to-end. Requires: architect skill. Optional: founder-mode


Product Builder

End-to-end product builder with founder mindset. Orchestrates 9 phases: idea evaluation (via architect) -> architecture -> TDD planning -> implementation -> E2E testing -> verification -> post-run analysis -> GTM launch kit (README, pitch, tweet thread, Product Hunt draft, CHANGELOG). Three human gates for Go/No-Go decisions. Crash recovery via state file. Supports new ideas and existing projects. The most ambitious skill in the marketplace — composes architect, learn-by-mistake, and post-run-analysis into a single pipeline. Requires: architect skill. Optional: learn-by-mistake, post-run-analysis


Pipelines

Skills that orchestrate other skills. One command in, finished artifact out.

Describe a demo in plain English, get a recorded video. Discovers your app's screens via static analysis, generates a test script, executes it with maestro (mobile) or webreel (web), and outputs screenshots + video + QA report. Requires: maestro or webreel

Auto-generate and run E2E tests from static codebase analysis. Scans project structure, identifies testable flows, generates scripts, executes them, reports results. Pairs with mobile-design-system for design-informed test generation. Requires: maestro CLI


Testing

Comprehensive Maestro E2E testing for React Native/Expo. 11 core patterns including auth pre-flight, adaptive flows, optimistic update verification, and OTP testing. iOS/Android gotchas, CI/CD with Maestro Cloud, and MCP server integration. The test execution engine for demo-pipeline and e2e-pipeline. Requires: maestro CLI, Java 17+


Recording & Automation

Record scripted browser demos from a JSON config. Cursor animation, keystroke HUD overlays, sound effects. Outputs MP4, GIF, or WebM. Custom themes, viewport presets, shared step sequences, and watch mode. Used by demo-pipeline for web recordings. Requires: webreel (npm)

Browser automation built for AI agents. Navigate, interact with @e refs, fill forms, upload files, drag-and-drop, execute JavaScript, and record video. 6 reference docs and 3 ready-to-use templates. Requires: infsh CLI (inference.sh)

Record browser interactions as video using Playwright. Cursor highlighting, click ripple effects, cookie banner dismissal, and window scaling patterns. Requires: Playwright


Design

Comprehensive mobile UI/UX design system covering React Native, iOS, and Android. Touch targets, gesture handling, haptic feedback, safe areas, accessibility, dark mode, and platform-specific patterns. Pairs with e2e-pipeline: design defines the rules, tests verify them. Requires: React Native or native platform

Redesign mobile app UI to feel unmistakably iOS-native with modern Apple-like glass materials (translucency, blur, depth). SF Pro typography, vibrancy layers, and material system. Requires: React Native or SwiftUI

Create App Store and Google Play screenshots with exact platform specs. Covers iOS/Android dimensions, gallery ordering, device mockups, captions, preview videos, localization, and A/B testing. Requires: None

Iterative UI/UX polishing workflow for web applications. Achieve Stripe-level visual polish through multiple passes — each iteration adds incremental improvements that compound dramatically. Requires: None


Documents

Professional PDF reports and presentations from structured content. Multi-page layouts, charts, branding, export-ready formatting. Turns architect evaluations and council advisories into polished deliverables — the final stage of the pipeline. Requires: None


Development

Create developer YouTube content, technical screencasts, and video tutorials. Covers tutorial structure, screen recording setup, live coding, shorts vs long-form, thumbnails, and SEO for technical content. Requires: None


How full-eval Orchestrates Everything

full-eval
  |
  |---> architect (6 evaluators in parallel)
  |       |-- Market Analyst
  |       |-- Technical Architect
  |       |-- Risk Assessor
  |       |-- Impact Evaluator
  |       |-- Resource Strategist
  |       |-- Innovation Scout
  |       └── Synthesizer --> Opportunity Score + Value Matrix
  |
  |---> council (8 advisors in parallel)
  |       |-- Finance       |-- Creativity
  |       |-- Health        |-- Productivity
  |       |-- Career        |-- Social Dynamics
  |       |-- Relationships |-- Personal Growth
  |       └── Synthesizer --> Unified Life Advisory
  |
  |---> Meta-synthesis (cross-system conflict resolution)
  |
  └──-> pdf-presentation --> Polished PDF Report

14 specialist agents. 2 synthesizers. 1 meta-synthesis. 1 PDF. One command.

The same multi-agent orchestration pattern powers hackathon-jury (3-7 jurors), vc-evaluator (3-5 investors), and oss-readiness (5 auditors) — each researching real people and evaluating from their genuine perspective.


How They Compose

Skills invoke other skills, chain outputs, and form pipelines.

idea-lab ──► architect (deep evaluation per idea)
         ──► founder-mode (build the winning idea, optional)
         ──► outputs: ranked dashboard + scenario recommendations

founder-mode ──► architect (idea evaluation)
             ──► learn-by-mistake (error memory, optional)
             ──► post-run-analysis (phase 7, optional)
             ──► outputs: working product + GTM launch kit

full-eval ──► architect + council + pdf-presentation
              (the complete evaluation pipeline)

demo-pipeline ──► maestro-mobile-testing (mobile apps)
               ──► webreel (web apps)
               ──► outputs: screenshots + video + QA report

e2e-pipeline ──► maestro-mobile-testing (test generation + execution)
             ──► mobile-design-system (design-informed test patterns)

architect ──► pdf-presentation (evaluation → polished PDF)
council   ──► pdf-presentation (advisory → formatted deliverable)

Any skill works standalone. The composition is optional but powerful.


Install

From the monorepo (Claude Code)

/install JuanMarchetto/agent-skills:skills/<skill-name>

Pick individual skills

/install JuanMarchetto/agent-skills:skills/learn-by-mistake
/install JuanMarchetto/agent-skills:skills/post-run-analysis
/install JuanMarchetto/agent-skills:skills/architecture-reviewer
/install JuanMarchetto/agent-skills:skills/hackathon-jury
/install JuanMarchetto/agent-skills:skills/full-eval
/install JuanMarchetto/agent-skills:skills/architect
/install JuanMarchetto/agent-skills:skills/vc-evaluator
/install JuanMarchetto/agent-skills:skills/oss-readiness
/install JuanMarchetto/agent-skills:skills/council
/install JuanMarchetto/agent-skills:skills/c-to-rust-migration
/install JuanMarchetto/agent-skills:skills/demo-pipeline
/install JuanMarchetto/agent-skills:skills/e2e-pipeline
/install JuanMarchetto/agent-skills:skills/maestro-mobile-testing
/install JuanMarchetto/agent-skills:skills/webreel
/install JuanMarchetto/agent-skills:skills/agent-browser
/install JuanMarchetto/agent-skills:skills/mobile-design-system
/install JuanMarchetto/agent-skills:skills/pdf-presentation
/install JuanMarchetto/agent-skills:skills/founder-mode
/install JuanMarchetto/agent-skills:skills/idea-lab
/install JuanMarchetto/agent-skills:skills/playwright-recording
/install JuanMarchetto/agent-skills:skills/ios-glass-ui
/install JuanMarchetto/agent-skills:skills/app-store-screenshots
/install JuanMarchetto/agent-skills:skills/ui-ux-polish
/install JuanMarchetto/agent-skills:skills/youtube-devrel
/install JuanMarchetto/agent-skills:skills/c-cpp-analyzer
/install JuanMarchetto/agent-skills:skills/rust-migration
/install JuanMarchetto/agent-skills:skills/migration-validator
/install JuanMarchetto/agent-skills:skills/noricum-dev

Manual Install

git clone https://github.com/JuanMarchetto/agent-skills.git
cp -r agent-skills/skills/<skill-name> ~/.claude/skills/<skill-name>

Compatibility

Tool Skills directory
Claude Code ~/.claude/skills/
Codex CLI ~/.agents/skills/
Gemini CLI ~/.agents/skills/
Cursor ~/.cursor/skills/
Windsurf ~/.agents/skills/

License

All skills are MIT licensed.

About

25+ skills for AI coding assistants — multi-agent panels, evaluation pipelines, testing, browser automation, and more

Topics

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages